Methods of manufacturing highly glycosylated glycoprotein hormones and their analogs

Abstract

Provided herein are methods for the separation of recombinant glycoprotein hormone from a biological sample. The methods disclosed herein comprise the steps of (a) adding the sample to a dye affinity chromatography medium equilibrated with a first buffer at a pH of greater than about 5, the medium comprising a solid matrix attached directly or indirectly to a dye ligand; (b) removing non-hormone contaminants by washing the medium with a second buffer at a total salt concentration of greater than about 0.6M and at a pH or greater than about 5, the second buffer comprising at least two different salts; and (c) eluting the hormone from the medium with a third buffer comprising at least 5% by volume of a polyol.

Description

[0001] METHODS OF MANUFACTURING HIGHLY GLYCOSYLATED GLYCOPROTEIN HORMONES AND THEIR ANALOGS

[0002] BACKGROUND

[0003]

[0001] Glycoprotein hormones, such as follicle-stimulating hormone (FSH), luteinizing hormone (LH), chorionic gonadotropin (CG) and thyroid-stimulating hormone (TSH), play important roles in reproductive and thyroid functions. These hormones are glycoproteins, modified by the addition of carbohydrate molecules called glycans through a process called glycosylation, which is critical for their folding, heterodimerization and biological activities. Such post-translational modification results in microheterogeneity or the presence of isoforms (glycoforms) with significant variations in the composition and structure of their carbohydrates.

[0004]

[0002] Equine chorionic gonadotropin (eCG) is a member of the glycoprotein hormone family which belongs to the large cystine knot growth factor superfamily (CKGF). This is the most heavily glycosylated glycoprotein hormone, with up to 35% of its mass consisting of carbohydrate chains determining its extremely long plasma half-life in a native form known as Pregnant Mare Serum Gonadotropin (PMSG). Only primates and equids produce long-acting chorionic gonadotropin in their placental cells (syncytiotrophoblast and differentiated cup cells, respectively). Importantly, equine placental cup cells produce glycans with both a2-3 and a2-6 linked sialic acid and N-acetyllactosamine repeats (3, 4), which is consistent with extremely high % of carbohydrate content in eCG obtained from serum of pregnant mares. eCG has dual activity, meaning it has both FSH (follicle-stimulating hormone) and LH (luteinizing hormone) activity in all species, except equids. This dual activity makes placental and recombinant eCG and their analogs extremely valuable tools in reproductive management in all veterinary species. In primates and equids, the chorionic epithelium expresses an LH-like molecule (chorionic gonadotrophin, CG) (3). Equine CG (eCG) selectively binds to LH receptors in equids. In all other domestic mammalian species, equine CG (eCG) shows an extraordinarily high FSH activity in addition to its LH activity ("dual activity"). Since its market launch, this has therefore gained considerable importance as a comparatively inexpensive FSH analogue, mainly for use in ruminants and pigs. In contrast to the human CG (hCG), which may be isolated non-invasively from the urine of pregnant women and is widely used as LH analogue, eCG must be extracted from the blood of pregnant donor mares, as eCG concentrations in urine are only minimal. Following reports of many deaths and suffering of donor mares associated with eCG / PMSG collection in South American and other countries, the current practice of eCG / PMSG production has given rise to increasing public criticism and outrage. This has recently led to calls for a general worldwide production ban of all PMSG and its combination products containing PMSG

[0005] (4), including combination of PMSG with urinary hCG, which also requires an urgent replacement with its recombinant counterpart.

[0006]

[0003] In contrast to other dual N- and O-linked glycosylated proteins such as erythropoietin (EPO), in eCG there are up to 12 O-linked carbohydrate chains in eCG versus 4-5 in human (h)CG

[0007] (5), and only 1 or 2 in EPO. In that respect, eCG and hCG are more like mucins, thyroglobulin and some antibodies. The exact location and structure of O-linked glycans in both eCG and hCG are not very well characterized and can vary depending on the cell type used to produce the protein and the culture conditions. N-glycosylation is better understood and requires the presence of asparagine (N) in a consensus tri-peptide NXT / S recognition sequence (where X is any amino acid except proline, T is threonine and S is serine). The site-occupancy and carbohydrate structure at each site of Asn-X-Thr / Ser orThr / Ser, in a high % of produced molecules is affected by various factors, including production conditions and CHO cell host expression of glycosylation enzymes, which may result in significantly different carbohydrate composition and structure profiles than found in native glycoprotein hormones.

[0008]

[0004] Thus, obtaining an optimal glycosylation for bioactivity and half-life of heavily recombinant eCG, super agonistic neo glycosylated analogs and similar glycoprotein hormones is very challenging due to the complexity and heterogeneity of the glycosylation process, and many factors involved in production of a recombinant glycoprotein with consistent and biologically relevant N- and O-linked glycans, which are optimally terminated with sialic acid residues without multiple terminal galactose residues. Such presence of terminal sialic residues due to charge repulsion eliminate binding of the penultimate galactose residues to the asialoglycoprotein receptor (ASGPR) and subsequent clearance by hepatocytes in the liver (Refs).

[0009] Thus, the critical goal of optimal manufacturing is to obtain glycoforms with no or no more than one exposed galactose in each N- or O-linked carbohydrate chain.

[0010]

[0005] All previous attempts to produce recombinant eCG matching half-life and bioactivity of natural eCG (PMSG) were highly unsuccessful. Recombinant eCG with a very short half-life was produced in insect Sf-9, yeast Pichia pastoris and CHO-cells, likely caused by the liver clearance of high-mannose glycoproteins and absence of the terminal sialic acids (Refs, personal communication). Single chain eCG with two covalently linked subunits resulted in high stability, but created production problems associated with multimers and aggregates, significant reduction of bioactivity and efficacy. There are no published reports of the carbohydrate structure of single-chain eCG analogs, which could be affected and limited by the position of 21- or 28- amino acid CTP-linker connecting C-terminus of eCG beta and N-terminus of eCG alpha subunit.

[0011] SUMMARY

[0012]

[0006] Provided herein are major novel modalities of optimal manufacturing of highly glycosylated, terminally sialylated glycoprotein hormones and combined production process, which was reduced to practice. Specific technical advantages and unexpected results of such combined process were previously not described in the prior art, in its entirety and specific parts are NOT obvious to a person skilled in the field. Multiple unexpected and surprising novel findings on optimal glycosylation of eCG and neo-glycosylated superagonist analogs provide a strong support for the non-obviousness of the entire combined manufacturing process and its specific parts.

[0013]

[0007] We have designed extension of plasma half-life with only one additional N-linked carbohydrate chain positioned within the N-terminus of alpha subunit, not interfering with its binding to the receptor and signaling. Trophogen's superagonist analog with 4 arginine substitutions whose potential epitopes have been masked by the additional N-linked carbohydrate chain have been designed and demonstrated to stimulate both FSH and LH / CG receptor as a dual activity gonadotropin, and no immunogenicity in tested animals.

[0014]

[0008] Using such novel eCG long-acting superagonist as a model, we have now devised a new manufacturing process resulting in a highly glycosylated product with both complex N-linked and O-linked carbohydrate chains terminated with sialic acid, which resulted in a long plasma halflife and maintained high in vivo bioactivity as tested in mice, rats and gilts. In particular, the presence of acidic isoforms (glycoforms) with pl between 2.5 and 4.5 indicating N-glycans with bi-, tri- and tetra-antennary complex type structures terminated predominantly with sialic acid resulting in a long-plasma half-life as tested in mice matching that of native eCG (PMSG). Thus, our new combined and refined manufacturing process assures presence of the desired N- and O- linked carbohydrate structures in the recombinant glycoprotein hormone.

[0009] The process described herein is not restricted to equine CG (eCG) but can be applied to all other recombinant glycoprotein hormones and their analogs from any vertebrate species. It is applicable not only to all highly glycosylated glycoprotein hormones, but also to human and primate CG molecules, but also to CGs from other equids such as donkey and zebras, which all contain both N- and O-linked carbohydrate chains.

[0015] DESCRIPTION OF THE DRAWINGS

[0016]

[0010] Fig. 1 is a schematic representation of Upstream processing.

[0017] [Oil] Fig. 2 is a schematic diagram of the Downstream processing scheme.

[0018]

[0012] Fig. 3 shows the chromatogram of the glycoprotein hormone purification using Blue-2.

[0019] The elution peak 2 containing glycoprotein hormone was analyzed by SDS-PAGE in reducing conditions and silver stained. It reveals —95% purity (Fig. 5). Two additional polishing steps include Mono Q and gel filtration chromatography (Figs. 4 and 5).

[0020]

[0013] Fig. 4 shows the chromatogram of monoQ purification after glycoprotein hormone eluted from Blue-2.

[0021]

[0014] Fig. 5: SDS-PAGE of the protein after each step purification. Lane 1 : glycoprotein hormone produced from CHO cell medium after dia-filtration. Lane 2: glycoprotein hormone after Blue purification eluted using glycerol and NaCI as described herein. Lane 3: glycoprotein hormone further purified by Mono Q. Lane 4: glycoprotein hormone further purified by Gel filtration. Lane 5: Gonal-F, a commercial glycoprotein hormone product as a control.

[0022] DETAILED DESCRIPTION

[0023] Upstream Processing of highly glycosylated glycoprotein hormones and their analogs

[0024]

[0015] New combination of the upstream process steps includes:

[0025] 1. Use of bicistronic expression vectors for high expression of eCG and its analogs allowing for separate amplification of individual subunits expression, high heterodimer production and potential reduction of free subunit levels.

[0026] 2. High CHO cell productivity of eCG and its analogs in CHO-DG44 cells (10-100 pg / cell / day), which is essential in bioreactor-based production and later processing eliminating major loses of highly glycosylated isoforms caused by other contaminating proteins and SFM media components.

[0027] 3. Adapting CHO cells to and using combination of specific carbohydrate precursors including N- Acetyl-mannosamine (ManNAc) and N-acetyl-galactosamine (GalNAc). Combination of 20 mM ManNAc and using 4 mM GalNAc resulted in more acidic isoform in IEF studies, lack of multimers or aggregates in SDS-PAGE and longer plasma half-life of eCG analog based on PK studies in mice. CHO cells prefer to utilize exogenous GalNAc, which also avoids competition for NeuAc precursor synthesis. Measuring galactose (Gal) level in SFM and adding if necessary exogenous Gal to SFM media to supply sufficient supply of the Gal precursor constitutes an additional promising precursor-based strategy.

[0028] 4. Highly controlled scale-up and bioreactor culture process using adherent CHO-DG44 cells with Cytopore 2 carriers providing excellent oxygenation control with limited shear stress a) Oxygen (O2) saturation in bioreactor - maintaining the saturation level at 45% during growth phase and 40% in perfusion / production phase to minimize chance of oxidative stress b) pH 7.0-7.2 c) Two steps reduction of temperature in bioreactor - the first to 34°C and the second to 32°C and using perfusion mode with constant monitoring of cell viability. In combination with other steps described this is more advanced than typical proliferation control strategy.

[0029] 5. Additional novel combination of methods enhancing glycosylation (as described separately in part V. below).

[0030]

[0016] Based on our testing and long-term experience with production of analogous optimally glycosylated and sialylated glycoprotein hormone superagonists (hTSH, hFSH, bFSH, hLH) high cell viability (>90%), oxygenation (40-45% saturation) and two-step lowering temperature to 32°C in bioreactor runs are essential for production of highly sialylated and glycosylated eCG analogs.

[0031]

[0017] To enhance glycosylation, eCG analog production was performed in the presence of carbohydrate precursors (substrates) including N-Acetylomannosamine (ManNAc) and N- acetylogalactosamine (GalNAc), which in combination with described above production conditions resulted in a generation of recombinant eCG analog with dominance of highly acidic isoforms with pl between 2.5 and 4.5, previously never achieved with any recombinant eCG preparations. Such dominance of acidic isoforms produced in bioreactor using enhanced culture process was surprising and related to the number of terminal sialic acid residues in both N- linked and multiple O-linked carbohydrate chains approaching acidic pl profile of natural eCG (PMSG), which is highly variable, batch-to-batch dependent with significant % of eCG dimers and other aggregates.

[0032] Upstream Methods

[0033] Production in bioreactor(s)

[0034]

[0018] We have used adherent CHO-DG44 cells stably transfected with engineered two bicistronic vectors, separate for each subunit. This allowed independent amplification of each subunit construct and optimization of production preferring eCG heterodimers rather than heterodimers mixed with excess of one subunit. Adherent CHO cells were selected based on our testing indicating generally higher specific per cell productivity of glycoprotein hormones than obtained using suspension cells, as well as higher stability and better control in longer bioreactor runs (more than 7-10). We have resolved the scalability issue with adherent cells using a new generation of microporous carriers (Cytopore 2 with crosslinked dextran) similar to previously tested in lab scale production at Trophogen, and earlier in the large scale production of recombinant TSH (Thyrogen)( {Cole, 1993 #80}), but allowing now for much better oxygenation control and reduced shear stress than all previous microcarriers, which are even more critical for highly glycosylated recombinant such as eCG and its analogs.

[0035]

[0019] We have used two different 5L bioreactor types in several runs using eCG analog producing CHO adherent cells: (1) packed-bed bioreactor CelliGen Plus from New Brunswick Scientific / Eppendorf using polyester disks and (2) Sartorius disposable plastic vessel stirred-tank bioreactor with a new generation of microporous carriers (Cytopore2), which allow for much than all previous carriers oxygenation and reduced shear stress. We have performed manufacturing scale-up from 1L-5L to 20L bioreactor and ensured that the product can be produced consistently at a larger scale. This process involved testing and validating the manufacturing process, ensuring that the product meets quality standards. In a larger GMP scale, an engineering run is an additional test production of the bioreactor process before full- scale commercial production in 100 or 200L.

[0036]

[0020] We have produced and purified 6 new batches of the lead eCG analog produced using Research Cell Bank (RCB) of CHO-DG44-eCG analog cells was tested in 3 different production systems: (1) Roller bottles to create cell inoculum for bioreactor runs; (2) Eppendorf / New Brunswick fibrous bed bioreactor using single-use disposable 5L vessel from Eppendorf-New Brunswick Scientific (2 runs, 3 batches: Al, A2, A3); and (3) Stirred-tank bioreactor (2L) run using microporous carriers

[0037]

[0021] CHO-DG44 cells producing eCG analog were inoculated from frozen vials into IMDM medium with 10% FBS and 400nM MTX. Cultures were scaled up from T-flasks after adaptation to Sigma Excell CHO CD medium to 15x 1700cm2extended surface roller bottles to obtain a minimum of 1.75X109total viable cells for bioreactor seeding. Approximately 2.0xl09cells were inoculated into a 5L-Eppendorf disposable bioreactor (3.75 L working volume) vessel with 150 g packed bed Fibra-cel disks. 350mL of complete culture medium (FBS of varying cone., 4mM L- glut, 400nM MTX) was perfused once daily from 0.1 to 1.2 bioreactor volume. Perfusion medium was changed on day 4 from 1% FBS to 0.1% FBS. 2 mM N-acetylmanosamine (Sigma) was used starting with partial harvest on day 7. Total eCG-T production in a single 10-days run was ~8mg. eCG levels were measured daily using eCG ELISA.

[0038]

[0022] Glucose and lactate levels were monitored daily using glucose and lactate meters via samples from perfused media. Lactate levels increased over time, and starting on Day 6, glucose levels slowly dropped. Culture mode - glucose was kept above 1.5 g / L and lactate below 1.5 g / L (16.65 mM). Dissolved oxygen was kept at 45% saturation. Temperature was initially maintained at 37°C and later lowered to 34°C and 32°C degrees; agitation at 100 rpm. Foaming was mild to moderate but kept in check with anti-foam agent and by decreasing sparge flow, increasing overlay air flow and antifoam agent as necessary. The pH 7.0-7.2 was maintained using a two-gas mixing system (Air and CO2) as well as sodium bicarbonate buffer in the culture media. Medium harvested from bioreactor was run in ultrafiltration mode on a Tangential Flow Filtration (TFF) skid with a PALL membrane cassette (0.1m2) with a lOkD MWCO and concentrated approximately lOx to reduce purification sample load. Concentrated harvest medium was then run on the TFF skid in Diafi Itration mode, and 7 volumes of media was exchanged into load buffer (25mM Sodium Phosphate). Sample load was then filtered (0.45pm PES) as a column guard before purification.

[0039]

[0023] For the Stirred-tank bioreactor (2L) run using microporous carriers (Cytopore2, GE) (3 batches: Bl-3), previously developed CHO-DG44 cells producing eCG analog were maintained as adherent cultures using microporous carriers. The working volume was 2-liters and contained approximately 2 grams of carriers per liter giving a total surface of 4.4 m2. The run was conducted over a 34-day period including monitoring of O2, pH and glucose concentrations, with perfusion being initiated on Day 12. Process control appears to have been suitable for all clones tested. Metabolite profiles were typical for CHO-DG44 cultures. The CD DG44 medium used was functioning very well and the viability of cells remained stable (viability of all tested clones was between 90-100%) as evidenced by a lack of non-viable cells in the culture supernatant. The endotoxin level was less than lEU / mL.

[0040]

[0024] Based on testing in stirred-tank bioreactor we increased the initial cell inoculum by factor of at least 5 and the microcarrier concentration by a factor of 3 as calculated per bioreactor vessel volume. This was modeled in spinner flasks prior to the initiation of a large volume perfusion run. The attachment of the cells to the microcarrier was monitored in daily media samples. The cells were allowed to expand for 7-10 days and the bioreactor was operated in a batch mode prior to the initiation of perfusion. Previously tested serum free media formulations from Sigma and Lonza allow maintaining high cell viability (>92%) and microcarrier attachment as evidenced by a lack of non-viable cells in the culture supernatant. Glucose and lactate levels is monitored daily using glucose and lactate meters via samples from perfused media. Glucose levels are maintained above 1.5 g / L, lactate below 1.5 g / L and dissolved oxygen was kept at 45% saturation. The supernatant collected during perfusion was collected, concentrated ~10 times using TFF with a PALL membrane cassette with a lOkD molecular weight cut-off and processed through a multi-step purification described below. The fractions collected from each production and purifications step are analyzed by Coomassie-Stained SDS-PAGE, and by Western Blots, as well as IEF followed by Western Blotting.

[0041]

[0025] We have tested carbohydrate precursors at various concentrations and found that combination of 20 mM N-Acetyl-mannosamine (ManNAc) and using 4 mM N-acetyl- galactosamine (GalNAc) is increasing the range of acidic eCG analog isoforms and results in a longer plasma half-life after single subcutaneous injection as tested in mice, in comparison to PMSG and less sialylated eCG preparations. Thus, supplementation of the serum free media with combination of specific precursors, such as GalNAc and ManNAc as well as nutrients in serum free media, such as galactose, fucose, and sialic acid, can increase the incorporation of desired carbohydrates. Addition of carbohydrate precursors in combination with optimized bioreactor culture conditions such as pH, lower temperature (32°C and 34°C), and oxygen levels (40-45% saturation), can improve the glycosylation of eCG analog.

[0026] We have observed that two-step lowering temperature of bioreactor run improved control of pH, oxygenation and other parameters critical for glycosylation and product stability in conditioned media. It lowered glucose consumption, lactate production, increased glutamine and slightly reduced ammonia concentration. Such lowering of the temperature was performed in two steps, the first to 34°C in the phase of cell growth plateauing with the first phase of perfusion culture and the second at the time of viability decrease below 90% lowering to 32°C and extending perfusion run. Most of the bioreactor process variables were positively- or not- affected by lower temperature, except some decrease of the specific growth rate which could be considered as a positive factor in a long-term bioreactor production phase with perfusion, when much faster decrease of cell viability (below 90%) was observed in standard 37°C production runs. Lowering temperature in combination with keeping pH above 7.0 is not only maintaining high cell viability, allows maintaining steady cell density, but is also preventing increase of sialidases and other degradation enzymes. We have previously observed that lower sialy lation in hollow fiber bioreactor is associated with high level of sialidases (REF). Therefore, monitoring sialidase activity during production and in the conditioned media provides additional assurance that the current process is leading to highly sialylated product.

[0042] Assessment of glycan profile during production

[0043]

[0027] Testing reproducibility of glycosylation of recombinant eCG batches produced in highly controlled process is solving major issues associated with great variations of eCG / PMSG glycosylation. Based on our studies on eCG analogs the completeness of N-glycosylation in the alpha subunit and O-glycosylation at the 28-amino acid CTP of eCG beta was improved with specific carbohydrate precursors as described and tested using isoelectric focusing (IEF) followed by Western blotting (WB). Such testing is done in 3 phases: a) initial testing of glycan profile of eCG produced by different CHO clones using testing IEF-WB; b) testing media collected during various phases of bioreactor run with IEF-WB using antibodies specific for wild-type eCG beta subunit; and c) confirmation that pl range of eCG isoforms match the results obtained of the PK after single injection in mice, in comparison to PMSG and other batch controls.

[0044]

[0028] Figure 1 is a schematic representation of Upstream processing. The procedure starts with removing one vial from WCB with appropriate logging for traceability. The content of the vial is inoculated in 125 mL flask with 30 mL of OptiCHO serum free medium. The flask is incubated at 37°C, 7% CO2 on orbital shaker on revolutions settings of 120 rpm. Cell density is monitored and recorded daily until reaching density of 1.5x106 cells / mL and viability at least 90%. The content of the flask is transferred to a 500 mL flask with 125 mL of OptiCHO serum free medium. Incubation continues on an orbital shaker at 37°C, 7% CO2 with revolutions settings of 120 rpm. When the cell density reaches 2-3xl06cells / mL with viability at least 90% the "Seed" is ready for inoculation to bioreactor. The "Seed" is inoculated to bioreactor using aseptic procedure. The bioreactor is filled with 50% of the total working volume of OptiCHO or similar serum free medium. The bioreactor's program is set to maintain temperature of 37°C, pH 7.0 and agitation of 80 rpm. The cell density, cell viability and content of glucose, lactate glutamine and glutamate, are monitored and recorded 2x daily. The collection of media starts 2 days after the cell density reaches 2-3xl06cells / mL. Based on cell density dynamics further harvests are performed cyclically until cell viability drops below 90%. At that point the temperature is lowered to 34°C. After harvest, equal volumes of pre-warmed OptiCHO serum free medium are injected back to the vessel. We have monitored cell density and viability 2 times a day during the entire production run. Based on the harvesting schedule samples for analysis of the TR44701 were collected and analyzed daily to support harvesting decisions based on human FSH ELISA testing and IEF-WB procedure. When a decision is made based on cell density and viability data, conditioned medium harvest is performed. Harvested conditioned medium is cleared by a series of filters with successively smaller pore size. First filter in process is Sartopure GF+ with pore size of 0.65 pm followed by Sartopure 2 with pore size of 0.45 pm. The filtration is performed using peristaltic pump Watson Marlow 520di with pump head 505L and tubing Masterflex id 25 to maintain appropriate flow rate and pressure. Collected and filtered medium is pumped directly through the last filter to the 3 or 5 L Flexboy bags in which it is stored at 4°C until transferred to "Downstream process." Commercial scale up of recombinant glycoprotein hormone analogs upstream process involves gradual increase of production volume from 5L to 200L with confirmation of product quality at each step, with an engineering run as the test of production in the bioreactor before the final full-scale commercial cGMP production.

[0045] Downstream Processing of highly glycosylated glycoprotein hormones and their analogs

[0046] Introduction

[0029] Downstream processing includes all activities and procedures that turn cleared conditioned media collected from the cell cultures to the final product. It now includes up to nine steps as shown in the schematic diagram in Fig. 2. Purification is performed after dialysis / buffer exchange using Blue-Sepharose HP column. Additional two major purification steps: anion-exchange and size exclusion chromatography. To assure that the process is safe from the blue dye leakage from the first initial capture stage, all material entering following step and the final product solution is tested for the presence of blue dye to eliminate any associated safety concerns.

[0047]

[0030] Our strategy in downstream process includes a quick application of conditioned media at 4°C to the first capture column (Blue Sepharose) and proceeding to the second step ionexchange step as soon as possible, decreasing early processing time and limiting the exposure of glycosylated product to sialidases and other degradation enzymes and even slightly lower that 7 pH produced by CHO cells. Such immediate processing limits removal of sialic acid and allows for maximum sia lylation of purified product.

[0048]

[0031] The major steps of the Downstream processing scheme are described below.

[0049]

[0032] TFF - Dia-concentration 1, 2 and 3 Steps (TFF) are performed on Sartorius Sartocon pump and Sartocon 200 stainless steel filter holder with filter Hydrosart 10K installed in Laminar Flow Hood for the time of filtration to maintain highest possible control over surrounding environment to decrease exposure to potential bioburden to minimum. The concentration procedure stops when the volume of all bags is concentrated to the volume calculated according to the planned scale. When the volume in the system reaches the determined target volume, the tubing feed with buffer for next column is attached to the system.

[0050]

[0033] TFF - Dia-concentration 4. Tangential Flow Filtration is performed on a Sartorius Sartocon pump and Sartocon 200 stainless steel filter holder with filter Hydrosart 10K installed in Laminar Flow Hood. The step "TFF - concentration 4" is performed to decrease the volume of pooled fractions from SEC / gel filtration step to volume calculated for the adjustment to final formulation (see below).

[0051]

[0034] Small volume concentration. Size exclusion chromatography requires that volumes of injected analyte are less than 5% of the column volume (CV) to keep high resolution level. Current column used is TOSOH GSK3000SW 600 x 20 mm with CV of 213 mL. Therefore, maximal injection volume should be 10 mL. The volume of analyte collected from previous steps should be decreased appropriately not to exceed protein concentration of 30 mg / mL. As the method of choice centrifugal concentration using Centricon lOkD units were used.

[0052]

[0035] Filtration 0.22 mm step is performed to keep minimal level of particulate contamination including microorganisms, microcrystals, precipitates and aggregates. This step is designed to remove potential precipitation seeds in further concentration step, decrease bioburden and increase lifespan of the SEC column.

[0053] Methods - major purification steps

[0054]

[0036] 2A. Blue Sepharose - In recent decades, recombinant glycoprotein hormones expressed in mammalian cells were purified in most cases using immuno-affinity chromatography. For instance, in the biotech industry, Gonal-F, a recombinant glycoprotein hormone from Serono, is purified using an immuno-affinity step. However, glycoprotein hormone elution from an immuno-affinity column usually involves harsh eluting conditions such as pH 3 or pH 10 which can easily denature or dissociate glycoprotein hormone subunits resulting in a low recovery yield. Since various lots of antibody are different and since there is no satisfactory procedure to clean the antibody-coupled column, it makes the immuno-affinity method not practical in a large-scale production and purification process. Moreover, the immuno-affinity method is associated with an increased cost of production and additional difficulties raised by requirements of regulatory agencies (e.g., USFDA). Organon published a glycoprotein hormone purification protocol using hydrophobic interaction, Prometic Blue-2 A6XI dye affinity, ionexchange and gel filtration chromatography. However, a major contaminant protein from conditioned medium is co-eluted from the Blue-2 A A6X1 column resulting in a need for additional purification steps. A U.S. patent by Vitro Diagnostics, Inc. in 1999 (U.S. Patent No. 5,990,288) also describes methods using Prometic dye affinity to purify glycoprotein hormone and elute glycoprotein hormone with 0.6 M NaCI at pH 6.0. Similar to the method used by Organon, this method does not separate the major contaminant, alpha-fetoprotein.

[0055]

[0037] Since we were not able to separate the major contaminating protein from glycoprotein hormone using Prometic Blue-2 A6XL (or other dye affinity such as Yellow, Red, Orange, etc. from Prometic) we have tested Blue Fast Flow from Amersham. We have found that alphafetoprotein and other contaminants can be eluted from Blue Fast Flow using 2M KCI. However, only 5% glycoprotein hormone can be eluted in such conditions. This indicated to us that if we can elute glycoprotein hormone from this column after washing out the most contaminant proteins by 2M KCI, we will be able to purify glycoprotein hormone to 90% purity in one single step. To elute glycoprotein hormone from Amersham Blue Fast Flow, we have tested ethylene glycol or KSCN as reported for IFN. Both ethylene glycol and KSCN can elute glycoprotein hormone from Blue 10 Fast Flow very efficiently; however, the eluted glycoprotein hormone showed low bioactivity after desalting as tested using cAMP stimulation and comparison with immuno-affinity purified glycoprotein hormone. To solve this problem, we have tested many different conditions and found that 20% glycerol with 2M KCI can effectively elute glycoprotein hormone from Blue Fast Flow (Amersham). The conditioned medium from cell culture was first concentrated and dia-filtrated using binding buffer (25 mM sodium phosphate, pH 7.0.) and then passed through Blue Fast Flow equilibrated with the same binding buffer. The glycoprotein hormone bound column was first washed with washing buffer (25mM sodium phosphate, 2M KCI, pH 7.0) to remove most contaminant proteins where alphafeto protein is the major one, then was eluted with elution buffer (25mM sodium phosphate, 2M KCI, 20% glycerol, pH 7.0). After this one step purification, the eluted glycoprotein hormone is already more than 90% pure and the recovery yield is more than 90%. Thus, our novel procedure constitutes the most effective known method of single-step purification of glycoprotein hormone and glycoprotein hormone analogs, not involving immuno-affinity chromatography. In contrast other methods (Serono, Organon) use 4-5 step procedures to achieve 90% purity with much lower overall recovery (50-60%).

[0056]

[0038] A number of additional methods (alone and in combination) may be employed to obtain optimal sia lylated N- and O-linked PMSG-like glycosylation profile.

[0057]

[0039] Serum free media and production optimization, including combination of adjusting the pH to 7.0-7.4 range, two-step lowering temperature to 34°C and to 32°C, and the addition of GalNAc to the culture media to increase the level of O-glycosylation.

[0058]

[0040] CHO cells stably transfected with alpha 2-6 sialyltransferase. Standard CHO cells express only a2,3-sialyltransferase (St3ga 11) but not a2,6-sialyltransferase (St6gall). Overexpression of both a2-3 and a2-6 sialyltransferases in CHO cells can be combined with addition of carbohydrate precursors to the culture media, as well as overexpression of Ugp2 and Galntl enzymes and addition of GalNAc and Gal enhancing O-glycosylation.

[0041] Addition of sialidases inhibitors and other enzyme inhibitors to conditioned media is another method reducing all unwanted changes in the glycosylation pattern of the eCG product, which is affected by sialidases and other enzymes released by CHO cells.

[0059]

[0042] CHO cell line engineering - overexpression of other selected enzymes to generate multimeric N-acetyl galactosamine units in O-linked chains mimicking those found in PMSG produced naturally in horse / mare by chorionic girdle cells. Such larger size O-linked carbohydrate chains should lead to longer half-life. It has been demonstrated that a portion of natural eCG (PMSG) contains branched poly-(N-acetyl mannosamine) peripheral extensions in O- linked. Such elongations of 1 O-glycans are not produced in CHO cells. Thus, engineering CHO cells with selected enzymes (Ugp2 and Galntl) to generate branched multimeric N-acetyl galactosamine units result in eCG analog with even longer plasma half-life and in vivo bioactivity. Using site-specific integration in CHO cells mediated by CRISPR / Cas9 and homology directed DNA repair pathway result in engineering CHO cells with largely unaltered productivity and growth rate.

[0060]

[0043] Use of engineered CHO cells with selective reduction of CHO host cell proteins / glycoproteins do not reduce target glycoprotein hormone productivity and glycosylation; variables include the number of antennas and terminal sialy lation in N-linked sugars; the completeness of O-glycosylation; etc.). Elimination of selective host cell proteins without any effect on productivity of eCG analog product characteristics, minimizing the number of purification steps. Use of such modified cells leads to removal of impurities and contaminants, increasing purity and safety of the final product.

[0061]

[0044] In-silico redesign of the adjacent sequence to Ser orThr using AlphaFold2, NetOGIyc 4.0 and similar tools can be used to refine position of different amino acid residues, including prolines in the CTP sequence of eCG beta.

[0062]

[0045] CHO cells with selective knock-out of sialidases / neuraminidases using CRISPR / Cas9 technology may be used. Exemplary knock-outs may include Neuraminidase 1 and 2 (Neul and Neu2), but not sialidases known to be involved in signaling, adhesion and cell growth (Neu 3 and Neu4).

[0063]

[0046] Use of alternative cell lines: CHO cells are not the only cell line that can be used for protein production. Other cell lines, such as HEK293 cells, can produce proteins with different O- glycosylation patterns. In particular, the use of equine chorionic girdle cell lines should allow expression of eCG analog with PMSG-like O-glycosylation profile including multimeric N-acetyl galactosamine units in O-linked chains.

[0064] Example 1: Production and Purification of glycoprotein hormone -4401 Lot 3M.

[0065]

[0047] A CHO cell line stable expressing glycoprotein hormone was cultured in Celligen plus Bioreactor with Fibracell disks in perfusion mode. When the cells were near confluent on fiber cell disks, cell culture medium was replaced by chemically defined CHO medium from Signa(C- 0488) for production. The collected medium around 10 L containing 174000 III glycoprotein hormone (glycoprotein hormone ELISA) was filtered first to remove the cell debris and then applied to 10 K Hydrosa rt Ultrafiltration Cassettes (Sartorius) for ultrafiltration and diafiltration. During this process, the medium was concentrated 10 times, and the buffer was changed to Binding Buffer (25 mM Sodium Phosphate, pH 7.0). The IL concentrated medium was then ready to be applied to Blue-2 column. The chromatography column used for Blue-2 was XK 50 / 100 from Amersham Bioscience. It was packed with 150 ml Sepharose 6 Fast Flow (Amersham Bioscience) according to the manufacturer's recommendations at room temperature. The column was operated on AKTA Explorer HPLC System from Amersham Bioscience. It was equilibrated with binding buffer (25 mM Sodium Phosphate, pH 7.0) at 36 cm / h for 4 column volume. The IL concentrated sample was then applied to the column at 36 cm / h. During this process, 90% glycoprotein hormone was bound to the column. The flow-through glycoprotein hormone was near 18000 IU, just 10.3 % of total glycoprotein hormone sample. After being reapplied to the column, the flow-through has little glycoprotein hormone left. The unbound proteins were washed out by 2 column volume of binding buffer. The column was first eluted with elution buffer 1 (25 mM sodium phosphate, 2M KCI, pH 7.0) at the same speed. This elution only eluted less than 2% bound glycoprotein hormone but was very effective in removing most contaminating proteins such as fetoprotein like proteins. Finally, the bound glycoprotein hormone was eluted by elution buffer 2 (25 mM Sodium Phosphate, 2M KCI, 20% glycerol, pH 7.0). Adding 20% glycerol in elution buffer can effectively elute most bound glycoprotein hormone and results in at least 90% recovery.

[0066]

[0048] Based on our studies on purification of eCG analog it was recently found that the elution of eCG analog can also be achieved with only KCI gradient application (0-2M KCI; 0-100%) without use of 20% glycerol, which was essential for purification of other superagonists (FSH and TSH). Such a procedure without use of 20% glycerol plus 2M KCI is preferred for eCG analog, and limits additional buffer exchanges related to the presence of glycerol which may also interfere with monitoring of the eCG levels by ELISA.

[0067] Example 2: steps before loading on Blue Sepharose in small scale production.

[0068]

[0049] Produced with Sigma Excell CHO CD media with 20 mM N-acetyl mannosamine (ManNAc). Cell viability monitored viability to determine harvest time. Supernatant harvested by centrifugation. Purification: TFF membrane holder using 1 x 0.1m2Omega T-Series 10K MWCO (sanitized). Flush with 25mM Sodium Phosphate pH 7.0. Concentrate the sample 10X. Dia-filter using 7DV of 25mM Sodium Phosphate, pH 7.0. Recover any product from the membrane by recirculating a small volume of equilibration buffer through the membrane. Filter the concentrated sample through a 0.45pm filter. 100 mL Column packed with Blue Sepharose 6 FF was loaded with lOx dilution of concentrated samples in 25 mL sodium phosphate, pH 7.0. 3L elutions from 3 purifications pooled, concentrated, buffer exchanged into PBS, pH 7.3 to a final volume of ~80 ml, which was concentrated 2 times to a final 40 ml. All samples were 0.2 pm filtered.

[0069]

[0050] The presence of blue dye on the Blue Sepharose beads column is used to monitor the purification process. The dye is visible and can be used as an indicator of the progress of purification. By monitoring the appearance of the blue dye in the eluate, one can determine when the desired protein has been purified to a sufficient degree. The blue dye covalently attached to the Blue Sepharose matrix has a characteristic absorbance peak at a wavelength of around 595 nm. This absorbance can be used to quantify the amount of dye present in a sample or to monitor the purification process. During the purification process, the Blue Sepharose beads are typically washed with a buffer containing a lower concentration of salt or other competing ligands to remove any unbound proteins. The absorbance of the buffer solution is typically measured at 595 nm using a spectrophotometer. As the purification process progresses, the amount of bound blue dye on the Blue Sepharose beads decreases, resulting in a decrease in the absorbance of the buffer solution. This decrease in absorbance was used as an indicator of the progress of the purification process and helped to determine when the desired protein has been purified to a sufficient degree.

[0051] Anion-Exchange - For ion-exchange step we have evaluated Bio-Works (Uppsala, Sweden) IEX resins to perform the separation after the first step using a screening kit which included 1 and 5 ml pre-packed BabyBio columns with our Q, S, DEAE and TREN (multimode anion) resins. For eCG, FSH, TSH and their analogs a strong anion exchanger - Q Sepharose was best to remove contaminants remaining after the first capture step. It is possible that different serum free media or slightly modified conditions will require use of other ion-exchange columns, initially in a format of small, prepacked columns.

[0070]

[0052] Anion Exchange is performed to further purify intermediate product and remove potential murine RNA viruses and. Target glycoprotein hormone is excluded from binding to Q resin because of ionic strength of buffer 2A1. Collected flow-through will be purified further in the next steps.

[0071]

[0053] Gel Filtration / Size Exclusion Chromatography (SEC) - This is a final polishing step which eliminates possible multimers and aggregates. It is known that natural eCG (PMSG) can exist in a heterodimeric form, consisting of two subunits (alpha and eCG beta), or a form, consisting of two heterodimers. In addition to such forms, eCG can also form higher order multimers such as trimers, tetramers, and pentamers. These multimers are formed by the association of several heterodimeric units, and their formation is quite common in various PMSG preparations, as well as previously described single chain eCG preparations, and thought to be influenced by factors such as pH, ionic strength, and the presence of specific binding partners or unpurified media components. The different forms of eCG, including multimers, can have highly different biological activities and pharmacokinetic properties. It is not clear if immunoreactive multimers are not a factor contributing to PMSG very long elimination phase. For example, multimeric forms of eCG have been shown to have longer half-lives and altered biological activity compared to the heterodimeric form. The specific composition and distribution of eCG multimers can vary depending on the source and preparation of the hormone. SDS-PAGE and Western blotting showed significant amount (>20%) of eCG high molecular weight aggregates in PMSG control preparation. Such eCG aggregates detected in two different commercial PMSG preparations were not found in our eCG analog recombinant preparations.

[0072]

[0054] The Size Exclusion step was performed using Superdex 75, 200 or TOSOH GSK3000SW (600 x 20 mm) columns. The injection volume did not exceed 10 ml_. Based on scale of production multiple runs were performed. Column performance is tested using Molecular Weight Standard from BioRad). Acceptance criteria and system cleaning procedure are established. Collected fractions are pooled according to chromatogram and stored at -80° C.

[0073] Steps designed for viral clearance and sterility

[0074]

[0055] Anion Exchange HPLC step in addition to purification of the intermediate product serves as a method to remove potential murine RNA viruses. Target glycoprotein hormone is excluded from binding to Q resin because of ionic strength of the buffer used.

[0075]

[0056] Nanofiltration. This is the last step of "Purification procedure" before the "Formulation" step. The filter system Virosart® CPV shows efficient removal of both small non-enveloped viruses (20 nm) and large enveloped viruses (>50 nm) from biopharmaceutical feed stream by size exclusion. Virosart® CPV filters are designed, developed and manufactured in accordance with a ISO 9001 certified Quality Management System. Virosart® CPV filters have been validated for 4 loglO removal of small non-enveloped viruses. This step is carried out in a laminar flow hood.

[0076]

[0057] Viral Clearance - The manufacturing process is capable of removing or inactivating viruses that may be present in the starting material and that the final product is safe for use. The US Food and Drug Administration (FDA) recommends a minimum of 3-log (1000-fold) reduction for enveloped viruses and 4-log (10,000-fold) reduction for non-enveloped viruses. It is important to note that the required reduction factor is a minimum requirement, and the actual reduction achieved in the viral clearance study is likely higher. The goal of the study is to demonstrate that the manufacturing process is capable of removing or inactivating viruses, and any reduction in virus titer is considered beneficial for reducing the risk of viral contamination in the final product. Validation and evaluation are performed using a scale-down process with selected CRO according to ICH Q5A guidance recommending viral clearance validation when rodent cell lines (e.g. CHO) are used. Testing includes general viral clearance capacity with two viruses and the assessment of the robustness of the clearance capacity. It involves first demonstration of virus detection and then reduction of the virus titer in spiking studies.

[0077]

[0058] Sterility of produced batches was assured by using the membrane filtration method. Samples of the final product are filtered through a sterile membrane 0.22 mm filter. The filter is then transferred to a sterile growth medium and incubated to allow any possible microorganisms present in the sample to grow. If no growth is observed in the growth medium, it indicates that the product is sterile. Additional Mycoplasma testing was also performed and was negative. Aseptic filling is the final manufacturing step.

[0078] Final Product Formulation, Testing and Characterization

[0079]

[0059] Introduction - Our tested methods include optimization of formulation conditions, selection of appropriate stabilizing agents and lyoprotectants, use of advanced freeze-drying technologies and optimization of lyophilization cycle combine glycoprotein hormone isoforms and other properties. Optimization of formulation and delivery methods, the use of novel drug delivery systems, such as liposomes, nanoparticles, and hydrogels can be used to enhance stability, bioavailability, and targeted delivery of glycoproteins to specific tissues or cells. Such new methods of administration, delayed absorption and increased stability should address the requirements for largely superior veterinary and human glycoprotein hormones.

[0080] Formulation

[0081]

[0060] Liquid filing is performed as follows. Intermediate product from "Nanofiltration step" step is mixed inactive components listed in Table 2. Such formulation of glycoprotein hormone is based on previous experience with human FSH analogs TR4401 and TR4402, as well as bovine TR55601 that has been tested and is used in approved and marketed recombinant human FSH drug Gonal-F by Merck KGaA - EMD Serono. Appropriate volume of FFB3x buffer - considering target concentration of glycoprotein hormone and resulting in formulation.

[0082] Table 1. Buffers used in Purification Process and Formulation

[0083] Table 2. Liquid Formulation Components

[0084] 3. Stability

[0085]

[0061] Stability testing is carried out following ICH Final Guideline on Stability Testing of Biotechnological / Biological Products (Hubbard, 1996) and FDA Guidance for Industry Q1A (R2) Stability Testing of New Drug Substances. Testing includes appropriate stability-indicating test methods using validated stability storage capacity incubators, refrigerators (2-8°C), freezers (- 20°C), and low temperature freezers (-80°C).

[0086] 4. Product Release Testing (Identity, Purity, Potency)

[0087]

[0062] We have designed recombinant eCG lot release criteria guaranteeing lot-to-lot consistency combined with described above novel methods of expression and processing:

[0088] 1. SDS-PAGE and isoelectric Focusing (IEF) and IEF followed by Western blotting on purified and unpurified samples, respectively. Such quantitative IEF assessments of isoforms provide a reliable prediction of the in vivo bioactivity later tested in #3 and #4. Based on previous studies dominance of higher acidic isoforms in IEF is expected also to correlate with much slower elimination in PK study. However, the lack of isoforms with pl between 3.5 and 4.5, and prevalence of isoforms with higher pl than 4.5 strongly suggests not fully suitable manufacturing process.

[0089] 2. Carbohydrate content and structure mas spec (MALDI-TOF-MS) analysis on carbohydrates released from purified samples.

[0090] 3. Steelman-Pohley bioassay in 21-old Sprague-Dawley rats with and without hCG augmentation. It is critical that in addition to a standard Steelman-Pohley assay with 3 daily injections, bioactivity is tested with only a single injection of eCG, eCG + 3x daily injections of hCG, and control PMSG, considering the major effect of PK on a final bioactivity.

[0091] 4. PK in mice and gilts to account for any species dependent PK difference.

[0092] 5. Final clinical trials in gilts, cow, sheep etc. and humans depending on the hormones and intended species for clinical use.

[0093]

[0063] Assessment of in vivo bioactivity in rats is performed using Steelman-Pohley bioassay with or without hCG augmentation and eCG-induced increase of the ovarian weight in immature Sprague-Dawley rats. Not less than 40h and not more than 72h after the last injection, rats are euthanized. Both ovaries are removed and cleaned from any extraneous fluid and tissue, weighted immediately and the results calculated using the total mass of 2 ovaries of each animal as the response, in comparison to the International Standard of eCG or a reference preparation calibrated in International Units. Tested eCG preparations are quantified against the reference standard according to the published formulas. In contrast to other gonadotropins, eCG analog due to its dual activity at FSH and LH receptors can increase ovarian weight in rats without hCG augmentation.

[0064] PK profile of the eCG analog in comparison to PMSG is performed in male mice by sc administration of a single dose of 10 pg per animal and 10 different blood collections times (5, 30, 60 min and 2, 6, 24, 48, 72, 96 and 120h) spanning both distribution and elimination phase. Fully purified analog preparation is characterized in vitro by RP-HPLC, validated eCG ELISA immunoassay, SDS-PAGE electrophoresis and IEF gel analysis, two robust invitro cAMP bioassays using 293-LHR and CHO-FSHR cell lines, as well as stimulation of progesterone production in MAIO mouse Leydig cells. The final purity of eCG analogs is expected to be greater than 95%. Cumulative recovery is expected to reach at least 45% with a total 50-fold purification.

[0094] V. Additional methods enhancing quality of the final product

[0095]

[0065] 1. Introduction O-linked glycosylation is a type of post-translational modification that involves the attachment of a carbohydrate molecule (usually N-acetylgalactosamine, GalNAc) to the hydroxyl group of a serine (Ser) or threonine (Thr) residue in a protein. The recognition sequence for O-linked glycosylation varies depending on the specific enzyme that catalyzes the glycosylation reaction. However, a common feature of O-linked glycosylation recognition sequences is the presence of a serine or threonine residue followed by one or more amino acids and then a proline residue. Proline residue is important because it helps to position the hydroxyl group of the serine or threonine residue in the proper orientation for glycosylation to occur. The specific amino acid residues in the recognition sequence can vary depending on the enzyme involved in the glycosylation reaction and the particular protein being modified. In some cases, the glycosylation recognition sequence may be located in a flexible region of the protein, allowing for the modification to occur without disrupting the protein's structure or function.

[0096]

[0066] Predicting O-linked glycosylation sites based on the amino acid sequence can be challenging due to the complex nature of the glycosylation process and the lack of a clear consensus sequence for O-linked glycosylation. However, there are several computational tools available that use machine learning algorithms to predict O-linked glycosylation sites based on the amino acid sequence. These tools generally use a combination of sequence-based features, such as the presence of certain amino acids or motifs, as well as physicochemical properties of the amino acids to predict potential O-linked glycosylation sites. One such tool is NetOGIyc 4.0, which uses a neural network-based approach to predict O-linked glycosylation sites. Other tools include YinOYang 1.2 and O-Glycbase 2.0. It is important to note that these tools are not always accurate and should be used in conjunction with experimental validation to confirm predicted O- linked glycosylation sites. NetOGIyc 4.0 or similar programs can be used to optimize eCG beta subunit ca boxy-terminal peptide (CTP) amino acid sequence for optimal and complete O- glycosylation ( / .e. presence of all 12 O-linked chains). The amino acid sequence of eCG CTP is derived from the carboxyterminal end of the beta subunit of eCG and consists of 28 amino acids, although the exact sequence can vary between species. Improving O-glycosylation in CHO cells can be a complex and iterative process, but can be achieved through a combination of all of the strategies detailed herein.

[0097] 2. Additional Methods (alone and in combination):

[0098]

[0067] Serum free media and production optimization, including combination of adjusting the pH to 7.0-7.4 range, two-step lowering temperature to 34°C and to 32°C, and the addition of GalNAc to the culture media to increase the level of O-glycosylation.

[0099]

[0068] CHO cells stably transfected with alpha 2-6 sialyltransferase. Standard CHO cells express only a2,3-sialyltransferase (St3gall) but not a2,6-sialyltransferase (St6gall) (6). Overexpression of both a2-3 and a2-6 sialyltransferases in CHO cells can be combined with addition of carbohydrate precursors to the culture media, as well as overexpression of Ugp2 and Galntl enzymes and addition of GalNAc and Gal enhancing O-glycosylation (7).

[0100]

[0069] Addition of sialidase inhibitors and other enzyme inhibitors to conditioned media is another method reducing all unwanted changes in the glycosylation pattern of the eCG product, which is affected by sialidases and other enzymes released by CHO cells.

[0101]

[0070] CHO cell line engineering - overexpression of other selected enzymes to generate multimeric N-acetylgalactosamine units in O-linked chains mimicking those found in PMSG produced naturally in horse / mare by chorionic girdle cells. Such larger size O-linked carbohydrate chains should lead to longer half-life. It has been demonstrated that a portion of natural eCG (PMSG) contains branched poly-(N-acetyl mannosamine) peripheral extensions in O- linked. Such extensions elongations of 1 O-glycans are not produced in CHO cells. Thus, engineering CHO cells with selected enzymes (Ugp2 and Galntl) to generate branched multimeric N-acetyl galactosamine units should result in eCG analog with even longer plasma half-life and in vivo bioactivity. Using site-specific integration in CHO cells mediated by CRISPR / Cas9 and homology directed DNA repair pathway should result in engineering CHO cells with largely unaltered productivity and growth rate.

[0102]

[0071] Use of engineered CHO cells with selective reduction of CHO host cell proteins / glycoproteins hypothesized NOT to reduce target glycoprotein hormone productivity and glycosylation (focus on number of antennas and terminal sialylation in N-linked sugars; completeness of O-glycosylation, etc.). Elimination of selective host cell proteins without any effect on productivity of eCG analog product characteristics minimizing the number of purification steps. Use of such modified cells should lead to removal of impurities and contaminants, increasing purity and safety of the final product.

[0103]

[0072] In-silico redesign of the adjacent sequence to Ser orThr using AlphaFold2, NetOGIyc 4.0 and similar tools to refine position of different amino acid residues, including prolines in the CTP sequence of eCG beta.

[0104]

[0073] CHO cells with selective knock-out of sialidases / neuraminidases using CRISPR / Cas9 technology. Such knock-outs may include Neuraminidase 1 and 2 (Neul and Neu2), but not sialidases known to be involved in signaling, adhesion and cell growth (Neu 3 and Neu4)

[0074] Use of alternative cell lines: CHO cells are not the only cell line that can be used for protein production. Other cell lines, such as HEK293 cells, can produce proteins with different O- glycosylation patterns. In particular, the use of equine chorionic girdle cell lines should allow expression of eCG analog with PMSG-like O-glycosylation profile including, multimeric N- acetylgalactosamine units in O-linked chains. Other alternatives include trophoblast, placental, and amniotic cell lines.

Claims

We claim:

1. A method for the separation of recombinant glycoprotein hormone from a sample, comprising the steps of:- adding the sample to a dye affinity chromatography medium equilibrated with a first buffer at a pH of greater than about 5, the medium comprising a solid matrix attached directly or indirectly to a dye ligand;- removing non-hormone contaminants by washing the medium with a second buffer at a total salt concentration of greater than about 0.6M and at a pH or greater than about 5, the second buffer comprising at least two different salts; and- eluting the hormone from the medium with a third buffer comprising at least 5% by volume of a polyol.

2. The method of claim 1, wherein the dye ligand is Cibacron Blue 3G.

3. The method of claim 1, wherein the dye ligand is Orange 2 or Orange 3.

4. The method of claim 1, wherein the dye ligand is Red 2 or Red 3.

5. The method of claim 1, wherein the first buffer comprises a phosphate or a Tris salt and has a total salt concentration of less than about 0.2 M6. The method of claim 5, wherein the phosphate is sodium phosphate at a salt concentration of less than about 0.1 M.

7. The method of claim 6, wherein the salt concentration is about 25 mM.

8. The method of claim 1, wherein the at least two different salts of the second buffer comprise a phosphate and a halide salt.

9. The method of claim 8, wherein the halide salt is at a salt concentration of greater than about 1 M.

10. The method of claim 9, wherein the halide salt is potassium chloride at a salt concentration of about 2 M.

11. The method of claim 1, wherein the polyol is selected from the group consisting of glycerol, ethylene glycol and combinations thereof.

12. The method of claim 11, wherein the polyol is glycerol at a concentration of at least about 10% by volume.

13. The method of claim 12, wherein the glycerol is at a concentration of at least about 15% by volume.

14. The method of claim 13, wherein the glycerol is at a concentration of at least about 20% by volume.

15. The method of claim 1, wherein the third buffer further comprises a phosphate and a halide salt.

16. The method of claim 15, wherein the halide salt is at a salt concentration of greater than about 1 M.

17. The method of claim 16, wherein the halide salt is potassium chloride at a salt concentration of about 2 M.

18. The method of claim 1, wherein the solid matrix is selected from the group consisting of Sepharose, Sephadex, Sephacryl and combinations thereof.

19. The method of claim 1, wherein the matrix is attached directly to the dye ligand.

20. The method of claim 1, wherein the eluted glycoprotein hormone is at least about 85% pure.

21. The method of claim 1, wherein the eluted glycoprotein hormone is at least about 90% pure.

22. The method of claim 1, wherein the glycoprotein hormone is recovered in at least about 80% yield.

23. The method of claim 1, wherein the glycoprotein hormone is recovered in at least about 90% yield.

24. The method of claim 1, wherein the glycoprotein hormone is selected from the group consisting of human recombinant glycoprotein hormone, human urinary glycoprotein hormone, monkey glycoprotein hormone, bovine glycoprotein hormone, equine glycoprotein hormone, porcine glycoprotein hormone, ovine glycoprotein hormone, canine glycoprotein hormone, rat glycoprotein hormone, feline glycoprotein hormone and fish GTH-I.

25. A method for the separation of a cystine knot growth factor from a sample, comprising the steps of:- adding the sample to a dye affinity chromatography medium equilibrated with a first buffer at a salt concentration of less than about 300 mM at a pH of greater than about 5, the medium comprising a solid matrix attached directly or indirectly to a dye ligand;- removing contaminants by washing the medium with a second buffer comprising at least two different salts at a total salt concentration of greater than about 0.6 M; and- eluting the cystine knot growth factor from the medium with a third buffer at a salt concentration of greater than about 0.6 M and comprising a polyol.

26. The method of claim 25, wherein the cystine knot growth factor is selected from the group consisting of TSH, LH, hCG, eCG, GTH-II, Thyrostimulin, VEGF-A, VEGF-B, VEGF-C, VEGF-D, VEGF-like molecules, PLGF, PDGF-A, PDGF-B, PDGF-AB, NGF, BDNF, NT3, NT4, TGF- betal, activin A, activin B, activin AB, inhibin A, inhibin B, BMP-15 and a toxin with a cystine knot structure.

27. The method of claim 25, wherein the dye ligand is Cibacron Blue 3G.

28. The method of claim 25, wherein the at least two different salts of the second buffer comprise a phosphate and a halide salt.

29. The method of claim 28, wherein the phosphate is sodium phosphate at a salt concentration of less than about 0.1 M.

30. The method of claim 29, wherein the salt concentration is about 25 mM.

31. The method of claim 28, wherein the halide salt is potassium chloride at a salt concentration of greater than about 1 M.

32. The method of claim 31, wherein the salt concentration is about 2 M.

33. The method of claim 25, wherein the third buffer further comprises a phosphate and a halide salt.

34. The method of claim 33, wherein the phosphate is sodium phosphate at a salt concentration of about 25 mM.

35. The method of claim 33, wherein the halide salt is potassium chloride at a salt concentration of about 2 M.

36. The method of claim 25, wherein the polyol of the third buffer is selected from the group consisting of glycerol, ethylene glycol and combinations thereof.

37. The method of claim 36, wherein the polyol is glycerol at a concentration of at least about 20% by volume.

Images